All Posts

A FIX session is long-lived, stateful and ordered, so HA isn't about keeping two processes alive — it's about proving exactly one node owns the session in every failure mode. Working through that with HAProxy, Pacemaker, Corosync, quorum, fencing, persistent session state, and an etcd alternative.

A follow-up to the Cilium migration: what changes when ALB targets nodes instead of pods, how traffic actually flows through NodePort + kube-proxy replacement, and the operational surprises that surface after the migration is 'done'.

Why EKS's native VPC CNI hits a ceiling on pod density, and a step-by-step migration from VPC CNI to Cilium overlay networking on EKS 1.35.

Introducing my new portfolio website built with Hugo, featuring a monochrome design and integrated blog system for sharing DevOps and Kubernetes insights.

Essential security practices every DevOps engineer should implement when running Kubernetes clusters in production, from network policies to RBAC configuration.